Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x24f75a62fd0>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x24f75b1d588>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.0.1
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    inputs_real = tf.placeholder(tf.float32, (None,) + (image_width, image_height, image_channels), name='input_real')
    inputs_z = tf.placeholder(tf.float32, (None, z_dim), name='input_z')
    learning_rate = tf.placeholder(tf.float32, name='learning_rate')
    
    return (inputs_real, inputs_z, learning_rate)

  
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the generator, tensor logits of the generator).

def discriminator(images, reuse=False, alpha=0.1, keep_prob=0.8):
    """
    Create the discriminator network
    :param image: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function   
    with tf.variable_scope('discriminator', reuse=reuse):
        
        # Input layer is 28x28x3
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same', 
                                      kernel_initializer=tf.contrib.layers.xavier_initializer())
        x1 = tf.maximum(x1, x1*alpha)
        # 14x14x64
        
        x2 = tf.layers.conv2d(x1, 128, 5, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        x2 = tf.layers.batch_normalization(x2, training=True)
        x2 = tf.maximum(x2, x2*alpha)
        # 7x7x128
        
        # Flattening it!
        flat = tf.reshape(x2, (-1, 7*7*128))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        out = tf.nn.dropout(out, keep_prob)
        
        return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True, alpha=0.1):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function    
    with tf.variable_scope('generator', reuse=not is_train):
        # First fully connected layer
        x1 = tf.layers.dense(z, 7*7*256)
        
        # Reshape it to start the convolutional stack
        x1 = tf.reshape(x1, (-1, 7, 7, 256))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(x1, x1*alpha)
        # 7x7x256 print("x1:", x1.get_shape().as_list())
        
        x2 = tf.layers.conv2d_transpose(x1, 128, 5, strides=2, padding='same')
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(x2, x2*alpha)
        # 14x14x128 print("x2:", x2.get_shape().as_list())
        
        # Output layer
        logits = tf.layers.conv2d_transpose(x2, out_channel_dim, 5, strides=2, padding='same')
        # 28x28x3
        
        out = tf.tanh(logits)
                
        return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim, alpha=0.1, smooth=0.1, keep_prob=0.8):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    g_model = generator(input_z, out_channel_dim, alpha=alpha)
    d_model_real, d_logits_real = discriminator(input_real, alpha=alpha, keep_prob=keep_prob)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True, alpha=alpha, keep_prob=keep_prob)

    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real) * (1 - smooth)))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake

    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]

    # Optimize
    d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
    g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)
    
    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode, 
              print_every=20, show_every=500, alpha=0.1, smooth=0.1, keep_prob=0.8):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model        
    input_real, input_z, lr = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    
    d_loss, g_loss = model_loss(input_real, input_z, data_shape[3], alpha=alpha, smooth=smooth, keep_prob=keep_prob)
        
    d_opt, g_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    steps = 0
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                
                # Normalizing since they are between [-0.5, 0.5]
                batch_images = 2*batch_images
                
                steps += 1
                
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                _ = sess.run(d_opt, feed_dict={input_z: batch_z, input_real: batch_images})
                _ = sess.run(g_opt, feed_dict={input_z: batch_z})
                
                if steps % print_every == 0:
                    train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images})
                    train_loss_g = g_loss.eval({input_z: batch_z})
                    
                    print("Epoch: {}/{}\tStep: {}\t".format(epoch_i+1, epoch_count, steps),
                          "Discriminator Loss: {:.4f}\t".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                    
                if steps % show_every == 0:
                    show_generator_output(sess, 16, input_z, data_shape[3], data_image_mode)

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 64
z_dim = 100
learning_rate = 0.001
beta1 = 0.2
alpha = 0.1
smooth = 0.1
keep_prob = 0.8

print_every = 20
show_every = 460


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 20

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode, 
          print_every=print_every, show_every=show_every, alpha=alpha, smooth=smooth, keep_prob=keep_prob) # Some params added

Epoch: 1/20	Step: 20	 Discriminator Loss: 1.4691	 Generator Loss: 5.3325
Epoch: 1/20	Step: 40	 Discriminator Loss: 2.2402	 Generator Loss: 0.2259
Epoch: 1/20	Step: 60	 Discriminator Loss: 2.1768	 Generator Loss: 0.2238
Epoch: 1/20	Step: 80	 Discriminator Loss: 1.6170	 Generator Loss: 0.4271
Epoch: 1/20	Step: 100	 Discriminator Loss: 1.7064	 Generator Loss: 0.3738
Epoch: 1/20	Step: 120	 Discriminator Loss: 1.4612	 Generator Loss: 0.5094
Epoch: 1/20	Step: 140	 Discriminator Loss: 1.3578	 Generator Loss: 1.1369
Epoch: 1/20	Step: 160	 Discriminator Loss: 1.3256	 Generator Loss: 1.1134
Epoch: 1/20	Step: 180	 Discriminator Loss: 1.2966	 Generator Loss: 1.1221
Epoch: 1/20	Step: 200	 Discriminator Loss: 1.3214	 Generator Loss: 1.5033
Epoch: 1/20	Step: 220	 Discriminator Loss: 1.4910	 Generator Loss: 0.4650
Epoch: 1/20	Step: 240	 Discriminator Loss: 1.4677	 Generator Loss: 0.4639
Epoch: 1/20	Step: 260	 Discriminator Loss: 1.4151	 Generator Loss: 0.4828
Epoch: 1/20	Step: 280	 Discriminator Loss: 1.8088	 Generator Loss: 0.3126
Epoch: 1/20	Step: 300	 Discriminator Loss: 1.5054	 Generator Loss: 0.4506
Epoch: 1/20	Step: 320	 Discriminator Loss: 1.3119	 Generator Loss: 1.3251
Epoch: 1/20	Step: 340	 Discriminator Loss: 1.2747	 Generator Loss: 1.0792
Epoch: 1/20	Step: 360	 Discriminator Loss: 1.5489	 Generator Loss: 0.4208
Epoch: 1/20	Step: 380	 Discriminator Loss: 1.4272	 Generator Loss: 0.4968
Epoch: 1/20	Step: 400	 Discriminator Loss: 1.7868	 Generator Loss: 0.3081
Epoch: 1/20	Step: 420	 Discriminator Loss: 1.3544	 Generator Loss: 0.5458
Epoch: 1/20	Step: 440	 Discriminator Loss: 1.5090	 Generator Loss: 0.4374
Epoch: 1/20	Step: 460	 Discriminator Loss: 1.4415	 Generator Loss: 0.5092

Epoch: 1/20	Step: 480	 Discriminator Loss: 1.3851	 Generator Loss: 0.5268
Epoch: 1/20	Step: 500	 Discriminator Loss: 1.3488	 Generator Loss: 0.5913
Epoch: 1/20	Step: 520	 Discriminator Loss: 1.5034	 Generator Loss: 0.4788
Epoch: 1/20	Step: 540	 Discriminator Loss: 1.1985	 Generator Loss: 1.4353
Epoch: 1/20	Step: 560	 Discriminator Loss: 1.2341	 Generator Loss: 1.1486
Epoch: 1/20	Step: 580	 Discriminator Loss: 1.2506	 Generator Loss: 0.9658
Epoch: 1/20	Step: 600	 Discriminator Loss: 1.2564	 Generator Loss: 1.2420
Epoch: 1/20	Step: 620	 Discriminator Loss: 1.2916	 Generator Loss: 1.1729
Epoch: 1/20	Step: 640	 Discriminator Loss: 1.4195	 Generator Loss: 0.4972
Epoch: 1/20	Step: 660	 Discriminator Loss: 1.4207	 Generator Loss: 0.5135
Epoch: 1/20	Step: 680	 Discriminator Loss: 1.4053	 Generator Loss: 1.6091
Epoch: 1/20	Step: 700	 Discriminator Loss: 1.2526	 Generator Loss: 1.1218
Epoch: 1/20	Step: 720	 Discriminator Loss: 1.2640	 Generator Loss: 0.7717
Epoch: 1/20	Step: 740	 Discriminator Loss: 1.2094	 Generator Loss: 0.7423
Epoch: 1/20	Step: 760	 Discriminator Loss: 1.1396	 Generator Loss: 1.3993
Epoch: 1/20	Step: 780	 Discriminator Loss: 1.2567	 Generator Loss: 0.7552
Epoch: 1/20	Step: 800	 Discriminator Loss: 1.2070	 Generator Loss: 0.8553
Epoch: 1/20	Step: 820	 Discriminator Loss: 1.5154	 Generator Loss: 0.4310
Epoch: 1/20	Step: 840	 Discriminator Loss: 1.4541	 Generator Loss: 0.4714
Epoch: 1/20	Step: 860	 Discriminator Loss: 1.6656	 Generator Loss: 0.3719
Epoch: 1/20	Step: 880	 Discriminator Loss: 1.0910	 Generator Loss: 1.0265
Epoch: 1/20	Step: 900	 Discriminator Loss: 1.1431	 Generator Loss: 0.8356
Epoch: 1/20	Step: 920	 Discriminator Loss: 1.4985	 Generator Loss: 0.4389

Epoch: 2/20	Step: 940	 Discriminator Loss: 1.2312	 Generator Loss: 0.7410
Epoch: 2/20	Step: 960	 Discriminator Loss: 1.2459	 Generator Loss: 0.7076
Epoch: 2/20	Step: 980	 Discriminator Loss: 1.4920	 Generator Loss: 0.4407
Epoch: 2/20	Step: 1000	 Discriminator Loss: 1.1996	 Generator Loss: 1.6561
Epoch: 2/20	Step: 1020	 Discriminator Loss: 1.2072	 Generator Loss: 0.7458
Epoch: 2/20	Step: 1040	 Discriminator Loss: 1.5638	 Generator Loss: 0.3936
Epoch: 2/20	Step: 1060	 Discriminator Loss: 1.1214	 Generator Loss: 0.8453
Epoch: 2/20	Step: 1080	 Discriminator Loss: 1.2887	 Generator Loss: 1.2816
Epoch: 2/20	Step: 1100	 Discriminator Loss: 1.2055	 Generator Loss: 0.9153
Epoch: 2/20	Step: 1120	 Discriminator Loss: 1.4056	 Generator Loss: 0.5102
Epoch: 2/20	Step: 1140	 Discriminator Loss: 1.3818	 Generator Loss: 0.5127
Epoch: 2/20	Step: 1160	 Discriminator Loss: 1.1849	 Generator Loss: 1.1445
Epoch: 2/20	Step: 1180	 Discriminator Loss: 1.1358	 Generator Loss: 1.0783
Epoch: 2/20	Step: 1200	 Discriminator Loss: 1.5350	 Generator Loss: 0.4127
Epoch: 2/20	Step: 1220	 Discriminator Loss: 1.1553	 Generator Loss: 1.0995
Epoch: 2/20	Step: 1240	 Discriminator Loss: 1.5742	 Generator Loss: 0.4440
Epoch: 2/20	Step: 1260	 Discriminator Loss: 1.1943	 Generator Loss: 1.1268
Epoch: 2/20	Step: 1280	 Discriminator Loss: 1.2407	 Generator Loss: 0.7268
Epoch: 2/20	Step: 1300	 Discriminator Loss: 1.4549	 Generator Loss: 1.8820
Epoch: 2/20	Step: 1320	 Discriminator Loss: 1.2078	 Generator Loss: 0.8263
Epoch: 2/20	Step: 1340	 Discriminator Loss: 1.4533	 Generator Loss: 0.4577
Epoch: 2/20	Step: 1360	 Discriminator Loss: 1.4284	 Generator Loss: 0.4977
Epoch: 2/20	Step: 1380	 Discriminator Loss: 1.4347	 Generator Loss: 0.4770

Epoch: 2/20	Step: 1400	 Discriminator Loss: 1.0901	 Generator Loss: 0.8475
Epoch: 2/20	Step: 1420	 Discriminator Loss: 1.1411	 Generator Loss: 1.0580
Epoch: 2/20	Step: 1440	 Discriminator Loss: 1.0860	 Generator Loss: 0.8767
Epoch: 2/20	Step: 1460	 Discriminator Loss: 1.7793	 Generator Loss: 0.3063
Epoch: 2/20	Step: 1480	 Discriminator Loss: 1.4635	 Generator Loss: 0.4704
Epoch: 2/20	Step: 1500	 Discriminator Loss: 1.2076	 Generator Loss: 1.3912
Epoch: 2/20	Step: 1520	 Discriminator Loss: 1.0765	 Generator Loss: 0.9073
Epoch: 2/20	Step: 1540	 Discriminator Loss: 1.0876	 Generator Loss: 1.0389
Epoch: 2/20	Step: 1560	 Discriminator Loss: 1.2910	 Generator Loss: 0.6106
Epoch: 2/20	Step: 1580	 Discriminator Loss: 1.1866	 Generator Loss: 1.2778
Epoch: 2/20	Step: 1600	 Discriminator Loss: 1.0832	 Generator Loss: 1.4688
Epoch: 2/20	Step: 1620	 Discriminator Loss: 1.0858	 Generator Loss: 1.3121
Epoch: 2/20	Step: 1640	 Discriminator Loss: 1.1855	 Generator Loss: 0.7251
Epoch: 2/20	Step: 1660	 Discriminator Loss: 1.0854	 Generator Loss: 0.9120
Epoch: 2/20	Step: 1680	 Discriminator Loss: 1.5884	 Generator Loss: 2.4740
Epoch: 2/20	Step: 1700	 Discriminator Loss: 1.1845	 Generator Loss: 0.7084
Epoch: 2/20	Step: 1720	 Discriminator Loss: 1.1394	 Generator Loss: 0.8881
Epoch: 2/20	Step: 1740	 Discriminator Loss: 1.3966	 Generator Loss: 0.5006
Epoch: 2/20	Step: 1760	 Discriminator Loss: 1.2587	 Generator Loss: 0.6241
Epoch: 2/20	Step: 1780	 Discriminator Loss: 1.3587	 Generator Loss: 0.5209
Epoch: 2/20	Step: 1800	 Discriminator Loss: 1.1110	 Generator Loss: 0.9962
Epoch: 2/20	Step: 1820	 Discriminator Loss: 1.1082	 Generator Loss: 0.8751
Epoch: 2/20	Step: 1840	 Discriminator Loss: 1.2123	 Generator Loss: 0.6892

Epoch: 2/20	Step: 1860	 Discriminator Loss: 1.2400	 Generator Loss: 0.6224
Epoch: 3/20	Step: 1880	 Discriminator Loss: 1.3137	 Generator Loss: 0.6049
Epoch: 3/20	Step: 1900	 Discriminator Loss: 1.1014	 Generator Loss: 0.8674
Epoch: 3/20	Step: 1920	 Discriminator Loss: 1.3885	 Generator Loss: 0.5396
Epoch: 3/20	Step: 1940	 Discriminator Loss: 1.5374	 Generator Loss: 0.4388
Epoch: 3/20	Step: 1960	 Discriminator Loss: 1.1545	 Generator Loss: 0.7873
Epoch: 3/20	Step: 1980	 Discriminator Loss: 1.3159	 Generator Loss: 0.6010
Epoch: 3/20	Step: 2000	 Discriminator Loss: 1.1125	 Generator Loss: 0.7976
Epoch: 3/20	Step: 2020	 Discriminator Loss: 1.0300	 Generator Loss: 0.9429
Epoch: 3/20	Step: 2040	 Discriminator Loss: 1.4209	 Generator Loss: 0.4923
Epoch: 3/20	Step: 2060	 Discriminator Loss: 1.0640	 Generator Loss: 1.0163
Epoch: 3/20	Step: 2080	 Discriminator Loss: 0.9865	 Generator Loss: 1.4502
Epoch: 3/20	Step: 2100	 Discriminator Loss: 1.0902	 Generator Loss: 0.8475
Epoch: 3/20	Step: 2120	 Discriminator Loss: 1.1355	 Generator Loss: 0.7471
Epoch: 3/20	Step: 2140	 Discriminator Loss: 1.2871	 Generator Loss: 1.1303
Epoch: 3/20	Step: 2160	 Discriminator Loss: 1.0192	 Generator Loss: 1.5139
Epoch: 3/20	Step: 2180	 Discriminator Loss: 1.0124	 Generator Loss: 1.4331
Epoch: 3/20	Step: 2200	 Discriminator Loss: 1.6151	 Generator Loss: 0.4071
Epoch: 3/20	Step: 2220	 Discriminator Loss: 1.2092	 Generator Loss: 1.0454
Epoch: 3/20	Step: 2240	 Discriminator Loss: 1.2097	 Generator Loss: 0.8027
Epoch: 3/20	Step: 2260	 Discriminator Loss: 1.1427	 Generator Loss: 0.7168
Epoch: 3/20	Step: 2280	 Discriminator Loss: 1.1109	 Generator Loss: 0.7512
Epoch: 3/20	Step: 2300	 Discriminator Loss: 1.0719	 Generator Loss: 0.8616

Epoch: 3/20	Step: 2320	 Discriminator Loss: 1.0262	 Generator Loss: 0.8291
Epoch: 3/20	Step: 2340	 Discriminator Loss: 1.4541	 Generator Loss: 0.4458
Epoch: 3/20	Step: 2360	 Discriminator Loss: 1.1999	 Generator Loss: 0.7225
Epoch: 3/20	Step: 2380	 Discriminator Loss: 1.1371	 Generator Loss: 1.0868
Epoch: 3/20	Step: 2400	 Discriminator Loss: 1.0163	 Generator Loss: 0.9466
Epoch: 3/20	Step: 2420	 Discriminator Loss: 1.0711	 Generator Loss: 0.8432
Epoch: 3/20	Step: 2440	 Discriminator Loss: 1.6215	 Generator Loss: 0.4148
Epoch: 3/20	Step: 2460	 Discriminator Loss: 1.0334	 Generator Loss: 1.2487
Epoch: 3/20	Step: 2480	 Discriminator Loss: 1.7443	 Generator Loss: 0.3470
Epoch: 3/20	Step: 2500	 Discriminator Loss: 2.0639	 Generator Loss: 0.2521
Epoch: 3/20	Step: 2520	 Discriminator Loss: 1.0111	 Generator Loss: 0.9178
Epoch: 3/20	Step: 2540	 Discriminator Loss: 1.2433	 Generator Loss: 0.6621
Epoch: 3/20	Step: 2560	 Discriminator Loss: 1.3945	 Generator Loss: 0.5229
Epoch: 3/20	Step: 2580	 Discriminator Loss: 1.1412	 Generator Loss: 1.7042
Epoch: 3/20	Step: 2600	 Discriminator Loss: 0.9765	 Generator Loss: 1.0130
Epoch: 3/20	Step: 2620	 Discriminator Loss: 1.1512	 Generator Loss: 1.1353
Epoch: 3/20	Step: 2640	 Discriminator Loss: 1.1550	 Generator Loss: 0.9190
Epoch: 3/20	Step: 2660	 Discriminator Loss: 1.0667	 Generator Loss: 1.3208
Epoch: 3/20	Step: 2680	 Discriminator Loss: 1.6058	 Generator Loss: 0.3827
Epoch: 3/20	Step: 2700	 Discriminator Loss: 1.0999	 Generator Loss: 0.7855
Epoch: 3/20	Step: 2720	 Discriminator Loss: 2.2448	 Generator Loss: 3.3498
Epoch: 3/20	Step: 2740	 Discriminator Loss: 1.3541	 Generator Loss: 0.6378
Epoch: 3/20	Step: 2760	 Discriminator Loss: 1.2561	 Generator Loss: 0.6119

Epoch: 3/20	Step: 2780	 Discriminator Loss: 1.1520	 Generator Loss: 2.0531
Epoch: 3/20	Step: 2800	 Discriminator Loss: 0.8300	 Generator Loss: 1.2574
Epoch: 4/20	Step: 2820	 Discriminator Loss: 1.0339	 Generator Loss: 1.3447
Epoch: 4/20	Step: 2840	 Discriminator Loss: 1.2407	 Generator Loss: 0.6546
Epoch: 4/20	Step: 2860	 Discriminator Loss: 1.1402	 Generator Loss: 1.6943
Epoch: 4/20	Step: 2880	 Discriminator Loss: 0.9545	 Generator Loss: 2.0355
Epoch: 4/20	Step: 2900	 Discriminator Loss: 1.4160	 Generator Loss: 0.5365
Epoch: 4/20	Step: 2920	 Discriminator Loss: 1.1580	 Generator Loss: 0.6946
Epoch: 4/20	Step: 2940	 Discriminator Loss: 0.9177	 Generator Loss: 1.0353
Epoch: 4/20	Step: 2960	 Discriminator Loss: 0.9953	 Generator Loss: 1.6189
Epoch: 4/20	Step: 2980	 Discriminator Loss: 1.4721	 Generator Loss: 0.4677
Epoch: 4/20	Step: 3000	 Discriminator Loss: 1.7389	 Generator Loss: 0.3316
Epoch: 4/20	Step: 3020	 Discriminator Loss: 1.2143	 Generator Loss: 0.6502
Epoch: 4/20	Step: 3040	 Discriminator Loss: 1.1194	 Generator Loss: 0.7524
Epoch: 4/20	Step: 3060	 Discriminator Loss: 1.9430	 Generator Loss: 0.3513
Epoch: 4/20	Step: 3080	 Discriminator Loss: 1.0404	 Generator Loss: 0.8772
Epoch: 4/20	Step: 3100	 Discriminator Loss: 1.2178	 Generator Loss: 0.6449
Epoch: 4/20	Step: 3120	 Discriminator Loss: 1.3575	 Generator Loss: 0.5264
Epoch: 4/20	Step: 3140	 Discriminator Loss: 1.1856	 Generator Loss: 0.7574
Epoch: 4/20	Step: 3160	 Discriminator Loss: 1.0189	 Generator Loss: 1.6408
Epoch: 4/20	Step: 3180	 Discriminator Loss: 0.7789	 Generator Loss: 1.3146
Epoch: 4/20	Step: 3200	 Discriminator Loss: 1.5676	 Generator Loss: 0.4379
Epoch: 4/20	Step: 3220	 Discriminator Loss: 0.9841	 Generator Loss: 0.9774

Epoch: 4/20	Step: 3240	 Discriminator Loss: 1.4861	 Generator Loss: 0.4631
Epoch: 4/20	Step: 3260	 Discriminator Loss: 1.0949	 Generator Loss: 0.7951
Epoch: 4/20	Step: 3280	 Discriminator Loss: 0.9638	 Generator Loss: 1.1197
Epoch: 4/20	Step: 3300	 Discriminator Loss: 1.5093	 Generator Loss: 0.4633
Epoch: 4/20	Step: 3320	 Discriminator Loss: 1.0624	 Generator Loss: 0.9657
Epoch: 4/20	Step: 3340	 Discriminator Loss: 1.4392	 Generator Loss: 0.5504
Epoch: 4/20	Step: 3360	 Discriminator Loss: 0.9832	 Generator Loss: 0.9435
Epoch: 4/20	Step: 3380	 Discriminator Loss: 1.2356	 Generator Loss: 1.9402
Epoch: 4/20	Step: 3400	 Discriminator Loss: 2.1781	 Generator Loss: 0.2234
Epoch: 4/20	Step: 3420	 Discriminator Loss: 1.2632	 Generator Loss: 0.6795
Epoch: 4/20	Step: 3440	 Discriminator Loss: 0.8361	 Generator Loss: 1.3029
Epoch: 4/20	Step: 3460	 Discriminator Loss: 0.8077	 Generator Loss: 1.1646
Epoch: 4/20	Step: 3480	 Discriminator Loss: 1.0487	 Generator Loss: 1.0266
Epoch: 4/20	Step: 3500	 Discriminator Loss: 0.9223	 Generator Loss: 1.7897
Epoch: 4/20	Step: 3520	 Discriminator Loss: 0.9689	 Generator Loss: 2.0074
Epoch: 4/20	Step: 3540	 Discriminator Loss: 1.0322	 Generator Loss: 0.8783
Epoch: 4/20	Step: 3560	 Discriminator Loss: 0.7982	 Generator Loss: 1.3240
Epoch: 4/20	Step: 3580	 Discriminator Loss: 1.1083	 Generator Loss: 1.9981
Epoch: 4/20	Step: 3600	 Discriminator Loss: 0.8633	 Generator Loss: 1.1868
Epoch: 4/20	Step: 3620	 Discriminator Loss: 1.1276	 Generator Loss: 0.8084
Epoch: 4/20	Step: 3640	 Discriminator Loss: 1.0133	 Generator Loss: 0.8939
Epoch: 4/20	Step: 3660	 Discriminator Loss: 0.9489	 Generator Loss: 1.0718
Epoch: 4/20	Step: 3680	 Discriminator Loss: 1.6514	 Generator Loss: 0.4057

Epoch: 4/20	Step: 3700	 Discriminator Loss: 0.9216	 Generator Loss: 1.0017
Epoch: 4/20	Step: 3720	 Discriminator Loss: 0.8222	 Generator Loss: 1.4393
Epoch: 4/20	Step: 3740	 Discriminator Loss: 1.2702	 Generator Loss: 0.6320
Epoch: 5/20	Step: 3760	 Discriminator Loss: 0.8214	 Generator Loss: 1.3105
Epoch: 5/20	Step: 3780	 Discriminator Loss: 0.9958	 Generator Loss: 0.8891
Epoch: 5/20	Step: 3800	 Discriminator Loss: 1.1035	 Generator Loss: 2.4524
Epoch: 5/20	Step: 3820	 Discriminator Loss: 1.0580	 Generator Loss: 1.5447
Epoch: 5/20	Step: 3840	 Discriminator Loss: 1.0222	 Generator Loss: 1.6833
Epoch: 5/20	Step: 3860	 Discriminator Loss: 0.9975	 Generator Loss: 0.9030
Epoch: 5/20	Step: 3880	 Discriminator Loss: 0.8691	 Generator Loss: 1.2631
Epoch: 5/20	Step: 3900	 Discriminator Loss: 1.4126	 Generator Loss: 3.1306
Epoch: 5/20	Step: 3920	 Discriminator Loss: 0.7838	 Generator Loss: 1.8634
Epoch: 5/20	Step: 3940	 Discriminator Loss: 0.7759	 Generator Loss: 1.5148
Epoch: 5/20	Step: 3960	 Discriminator Loss: 0.6557	 Generator Loss: 1.9379
Epoch: 5/20	Step: 3980	 Discriminator Loss: 1.1567	 Generator Loss: 0.8268
Epoch: 5/20	Step: 4000	 Discriminator Loss: 1.0860	 Generator Loss: 2.3739
Epoch: 5/20	Step: 4020	 Discriminator Loss: 1.2672	 Generator Loss: 0.6486
Epoch: 5/20	Step: 4040	 Discriminator Loss: 1.0779	 Generator Loss: 0.8361
Epoch: 5/20	Step: 4060	 Discriminator Loss: 1.5407	 Generator Loss: 0.5002
Epoch: 5/20	Step: 4080	 Discriminator Loss: 0.8845	 Generator Loss: 2.1703
Epoch: 5/20	Step: 4100	 Discriminator Loss: 0.9920	 Generator Loss: 1.0436
Epoch: 5/20	Step: 4120	 Discriminator Loss: 1.4859	 Generator Loss: 0.5029
Epoch: 5/20	Step: 4140	 Discriminator Loss: 1.6879	 Generator Loss: 0.4796

Epoch: 5/20	Step: 4160	 Discriminator Loss: 1.1984	 Generator Loss: 1.8461
Epoch: 5/20	Step: 4180	 Discriminator Loss: 0.8731	 Generator Loss: 1.4707
Epoch: 5/20	Step: 4200	 Discriminator Loss: 0.8087	 Generator Loss: 1.2357
Epoch: 5/20	Step: 4220	 Discriminator Loss: 0.9516	 Generator Loss: 1.1177
Epoch: 5/20	Step: 4240	 Discriminator Loss: 0.9035	 Generator Loss: 1.0294
Epoch: 5/20	Step: 4260	 Discriminator Loss: 0.9872	 Generator Loss: 1.4838
Epoch: 5/20	Step: 4280	 Discriminator Loss: 0.8343	 Generator Loss: 1.2040
Epoch: 5/20	Step: 4300	 Discriminator Loss: 0.7118	 Generator Loss: 1.5226
Epoch: 5/20	Step: 4320	 Discriminator Loss: 1.2872	 Generator Loss: 0.6989
Epoch: 5/20	Step: 4340	 Discriminator Loss: 0.9000	 Generator Loss: 1.0674
Epoch: 5/20	Step: 4360	 Discriminator Loss: 0.6055	 Generator Loss: 2.3221
Epoch: 5/20	Step: 4380	 Discriminator Loss: 1.6918	 Generator Loss: 0.4607
Epoch: 5/20	Step: 4400	 Discriminator Loss: 1.7702	 Generator Loss: 0.3522
Epoch: 5/20	Step: 4420	 Discriminator Loss: 0.9935	 Generator Loss: 0.9938
Epoch: 5/20	Step: 4440	 Discriminator Loss: 0.9141	 Generator Loss: 1.0223
Epoch: 5/20	Step: 4460	 Discriminator Loss: 1.2020	 Generator Loss: 0.7105
Epoch: 5/20	Step: 4480	 Discriminator Loss: 0.6607	 Generator Loss: 1.7697
Epoch: 5/20	Step: 4500	 Discriminator Loss: 1.2738	 Generator Loss: 0.6982
Epoch: 5/20	Step: 4520	 Discriminator Loss: 1.2404	 Generator Loss: 3.1919
Epoch: 5/20	Step: 4540	 Discriminator Loss: 0.9821	 Generator Loss: 0.9391
Epoch: 5/20	Step: 4560	 Discriminator Loss: 1.5095	 Generator Loss: 0.5444
Epoch: 5/20	Step: 4580	 Discriminator Loss: 0.7818	 Generator Loss: 1.3560
Epoch: 5/20	Step: 4600	 Discriminator Loss: 0.8421	 Generator Loss: 2.1136

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 64
z_dim = 100
learning_rate = 0.001
beta1 = 0.1
alpha = 0.1
smooth = 0.1
keep_prob = 0.8

print_every = 20
show_every = 500


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 10

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode, 
          print_every=print_every, show_every=show_every, alpha=alpha, smooth=smooth, keep_prob=keep_prob) # Some params added

Epoch: 1/10	Step: 20	 Discriminator Loss: 5.4661	 Generator Loss: 0.0084
Epoch: 1/10	Step: 40	 Discriminator Loss: 2.5770	 Generator Loss: 0.1370
Epoch: 1/10	Step: 60	 Discriminator Loss: 2.5797	 Generator Loss: 0.1366
Epoch: 1/10	Step: 80	 Discriminator Loss: 2.1019	 Generator Loss: 0.2248
Epoch: 1/10	Step: 100	 Discriminator Loss: 1.5529	 Generator Loss: 0.4147
Epoch: 1/10	Step: 120	 Discriminator Loss: 1.7852	 Generator Loss: 0.2896
Epoch: 1/10	Step: 140	 Discriminator Loss: 0.4803	 Generator Loss: 2.3612
Epoch: 1/10	Step: 160	 Discriminator Loss: 2.9247	 Generator Loss: 4.4591
Epoch: 1/10	Step: 180	 Discriminator Loss: 0.7277	 Generator Loss: 1.7816
Epoch: 1/10	Step: 200	 Discriminator Loss: 0.6128	 Generator Loss: 2.2080
Epoch: 1/10	Step: 220	 Discriminator Loss: 1.3603	 Generator Loss: 0.9491
Epoch: 1/10	Step: 240	 Discriminator Loss: 2.2978	 Generator Loss: 2.5217
Epoch: 1/10	Step: 260	 Discriminator Loss: 3.0128	 Generator Loss: 2.8196
Epoch: 1/10	Step: 280	 Discriminator Loss: 1.2949	 Generator Loss: 0.8636
Epoch: 1/10	Step: 300	 Discriminator Loss: 1.5770	 Generator Loss: 1.7767
Epoch: 1/10	Step: 320	 Discriminator Loss: 1.2489	 Generator Loss: 1.4626
Epoch: 1/10	Step: 340	 Discriminator Loss: 1.6663	 Generator Loss: 0.3231
Epoch: 1/10	Step: 360	 Discriminator Loss: 1.4581	 Generator Loss: 0.4844
Epoch: 1/10	Step: 380	 Discriminator Loss: 1.2773	 Generator Loss: 0.6172
Epoch: 1/10	Step: 400	 Discriminator Loss: 1.3741	 Generator Loss: 0.5434
Epoch: 1/10	Step: 420	 Discriminator Loss: 1.3643	 Generator Loss: 0.5517
Epoch: 1/10	Step: 440	 Discriminator Loss: 1.4048	 Generator Loss: 0.8552
Epoch: 1/10	Step: 460	 Discriminator Loss: 1.4295	 Generator Loss: 1.3852
Epoch: 1/10	Step: 480	 Discriminator Loss: 1.4489	 Generator Loss: 1.4107
Epoch: 1/10	Step: 500	 Discriminator Loss: 1.4562	 Generator Loss: 0.4583

Epoch: 1/10	Step: 520	 Discriminator Loss: 1.5546	 Generator Loss: 0.3913
Epoch: 1/10	Step: 540	 Discriminator Loss: 1.2237	 Generator Loss: 0.6691
Epoch: 1/10	Step: 560	 Discriminator Loss: 0.6253	 Generator Loss: 3.5826
Epoch: 1/10	Step: 580	 Discriminator Loss: 1.8703	 Generator Loss: 1.9763
Epoch: 1/10	Step: 600	 Discriminator Loss: 1.4395	 Generator Loss: 0.5542
Epoch: 1/10	Step: 620	 Discriminator Loss: 1.5454	 Generator Loss: 0.4362
Epoch: 1/10	Step: 640	 Discriminator Loss: 1.4457	 Generator Loss: 0.4837
Epoch: 1/10	Step: 660	 Discriminator Loss: 0.9553	 Generator Loss: 1.0964
Epoch: 1/10	Step: 680	 Discriminator Loss: 1.4385	 Generator Loss: 0.5584
Epoch: 1/10	Step: 700	 Discriminator Loss: 1.6367	 Generator Loss: 0.9794
Epoch: 1/10	Step: 720	 Discriminator Loss: 1.4299	 Generator Loss: 0.4851
Epoch: 1/10	Step: 740	 Discriminator Loss: 1.5121	 Generator Loss: 0.9287
Epoch: 1/10	Step: 760	 Discriminator Loss: 1.3880	 Generator Loss: 0.9118
Epoch: 1/10	Step: 780	 Discriminator Loss: 1.3079	 Generator Loss: 0.5708
Epoch: 1/10	Step: 800	 Discriminator Loss: 1.4856	 Generator Loss: 0.7107
Epoch: 1/10	Step: 820	 Discriminator Loss: 1.5749	 Generator Loss: 0.6162
Epoch: 1/10	Step: 840	 Discriminator Loss: 1.3954	 Generator Loss: 0.8009
Epoch: 1/10	Step: 860	 Discriminator Loss: 1.4196	 Generator Loss: 0.5240
Epoch: 1/10	Step: 880	 Discriminator Loss: 0.9371	 Generator Loss: 1.4729
Epoch: 1/10	Step: 900	 Discriminator Loss: 1.5853	 Generator Loss: 0.4205
Epoch: 1/10	Step: 920	 Discriminator Loss: 1.0798	 Generator Loss: 1.0786
Epoch: 1/10	Step: 940	 Discriminator Loss: 1.3100	 Generator Loss: 0.5927
Epoch: 1/10	Step: 960	 Discriminator Loss: 1.5422	 Generator Loss: 0.4102
Epoch: 1/10	Step: 980	 Discriminator Loss: 1.2665	 Generator Loss: 0.8189
Epoch: 1/10	Step: 1000	 Discriminator Loss: 1.3769	 Generator Loss: 0.8511

Epoch: 1/10	Step: 1020	 Discriminator Loss: 1.5085	 Generator Loss: 0.5694
Epoch: 1/10	Step: 1040	 Discriminator Loss: 1.3655	 Generator Loss: 0.6492
Epoch: 1/10	Step: 1060	 Discriminator Loss: 1.1763	 Generator Loss: 0.8381
Epoch: 1/10	Step: 1080	 Discriminator Loss: 1.3001	 Generator Loss: 0.6099
Epoch: 1/10	Step: 1100	 Discriminator Loss: 1.1107	 Generator Loss: 0.8572
Epoch: 1/10	Step: 1120	 Discriminator Loss: 1.4239	 Generator Loss: 0.7509
Epoch: 1/10	Step: 1140	 Discriminator Loss: 1.5921	 Generator Loss: 1.3241
Epoch: 1/10	Step: 1160	 Discriminator Loss: 1.3632	 Generator Loss: 1.0100
Epoch: 1/10	Step: 1180	 Discriminator Loss: 1.5705	 Generator Loss: 0.9410
Epoch: 1/10	Step: 1200	 Discriminator Loss: 1.4953	 Generator Loss: 0.4998
Epoch: 1/10	Step: 1220	 Discriminator Loss: 1.3934	 Generator Loss: 0.6268
Epoch: 1/10	Step: 1240	 Discriminator Loss: 1.4358	 Generator Loss: 0.5646
Epoch: 1/10	Step: 1260	 Discriminator Loss: 1.5305	 Generator Loss: 1.3189
Epoch: 1/10	Step: 1280	 Discriminator Loss: 1.3640	 Generator Loss: 0.6674
Epoch: 1/10	Step: 1300	 Discriminator Loss: 1.2666	 Generator Loss: 0.7857
Epoch: 1/10	Step: 1320	 Discriminator Loss: 1.3176	 Generator Loss: 0.6872
Epoch: 1/10	Step: 1340	 Discriminator Loss: 1.3965	 Generator Loss: 0.7899
Epoch: 1/10	Step: 1360	 Discriminator Loss: 1.3806	 Generator Loss: 0.7788
Epoch: 1/10	Step: 1380	 Discriminator Loss: 1.4665	 Generator Loss: 0.6356
Epoch: 1/10	Step: 1400	 Discriminator Loss: 1.3992	 Generator Loss: 0.7567
Epoch: 1/10	Step: 1420	 Discriminator Loss: 1.3770	 Generator Loss: 0.6581
Epoch: 1/10	Step: 1440	 Discriminator Loss: 1.5425	 Generator Loss: 1.0063
Epoch: 1/10	Step: 1460	 Discriminator Loss: 1.7917	 Generator Loss: 1.5208
Epoch: 1/10	Step: 1480	 Discriminator Loss: 1.4126	 Generator Loss: 0.9285
Epoch: 1/10	Step: 1500	 Discriminator Loss: 1.4324	 Generator Loss: 0.5533

Epoch: 1/10	Step: 1520	 Discriminator Loss: 1.2647	 Generator Loss: 0.7504
Epoch: 1/10	Step: 1540	 Discriminator Loss: 1.3912	 Generator Loss: 1.0977
Epoch: 1/10	Step: 1560	 Discriminator Loss: 1.3896	 Generator Loss: 0.6159
Epoch: 1/10	Step: 1580	 Discriminator Loss: 1.3716	 Generator Loss: 0.9595
Epoch: 1/10	Step: 1600	 Discriminator Loss: 1.4306	 Generator Loss: 1.1088
Epoch: 1/10	Step: 1620	 Discriminator Loss: 1.2246	 Generator Loss: 0.7749
Epoch: 1/10	Step: 1640	 Discriminator Loss: 1.5013	 Generator Loss: 0.5041
Epoch: 1/10	Step: 1660	 Discriminator Loss: 1.3895	 Generator Loss: 0.7483
Epoch: 1/10	Step: 1680	 Discriminator Loss: 1.3641	 Generator Loss: 0.9026
Epoch: 1/10	Step: 1700	 Discriminator Loss: 1.3856	 Generator Loss: 0.8213
Epoch: 1/10	Step: 1720	 Discriminator Loss: 1.3417	 Generator Loss: 0.6931
Epoch: 1/10	Step: 1740	 Discriminator Loss: 1.3750	 Generator Loss: 0.6990
Epoch: 1/10	Step: 1760	 Discriminator Loss: 1.4313	 Generator Loss: 0.5918
Epoch: 1/10	Step: 1780	 Discriminator Loss: 1.5036	 Generator Loss: 0.8936
Epoch: 1/10	Step: 1800	 Discriminator Loss: 1.5499	 Generator Loss: 1.0587
Epoch: 1/10	Step: 1820	 Discriminator Loss: 1.3453	 Generator Loss: 0.7396
Epoch: 1/10	Step: 1840	 Discriminator Loss: 1.3726	 Generator Loss: 0.8474
Epoch: 1/10	Step: 1860	 Discriminator Loss: 1.5084	 Generator Loss: 0.9875
Epoch: 1/10	Step: 1880	 Discriminator Loss: 1.6865	 Generator Loss: 1.2708
Epoch: 1/10	Step: 1900	 Discriminator Loss: 2.1050	 Generator Loss: 1.4211
Epoch: 1/10	Step: 1920	 Discriminator Loss: 1.4312	 Generator Loss: 0.5649
Epoch: 1/10	Step: 1940	 Discriminator Loss: 1.3983	 Generator Loss: 0.6977
Epoch: 1/10	Step: 1960	 Discriminator Loss: 1.3395	 Generator Loss: 0.8011
Epoch: 1/10	Step: 1980	 Discriminator Loss: 1.4930	 Generator Loss: 1.0000
Epoch: 1/10	Step: 2000	 Discriminator Loss: 1.3949	 Generator Loss: 0.6610

Epoch: 1/10	Step: 2020	 Discriminator Loss: 1.4448	 Generator Loss: 0.5107
Epoch: 1/10	Step: 2040	 Discriminator Loss: 1.3363	 Generator Loss: 0.7236
Epoch: 1/10	Step: 2060	 Discriminator Loss: 1.3932	 Generator Loss: 0.9941
Epoch: 1/10	Step: 2080	 Discriminator Loss: 1.3252	 Generator Loss: 0.7347
Epoch: 1/10	Step: 2100	 Discriminator Loss: 1.3025	 Generator Loss: 0.8918
Epoch: 1/10	Step: 2120	 Discriminator Loss: 1.4396	 Generator Loss: 0.8136
Epoch: 1/10	Step: 2140	 Discriminator Loss: 1.5538	 Generator Loss: 0.4306
Epoch: 1/10	Step: 2160	 Discriminator Loss: 1.4363	 Generator Loss: 0.7477
Epoch: 1/10	Step: 2180	 Discriminator Loss: 1.5555	 Generator Loss: 0.4146
Epoch: 1/10	Step: 2200	 Discriminator Loss: 1.3266	 Generator Loss: 0.8278
Epoch: 1/10	Step: 2220	 Discriminator Loss: 1.7782	 Generator Loss: 1.4959
Epoch: 1/10	Step: 2240	 Discriminator Loss: 1.4652	 Generator Loss: 0.5898
Epoch: 1/10	Step: 2260	 Discriminator Loss: 1.4103	 Generator Loss: 0.5402
Epoch: 1/10	Step: 2280	 Discriminator Loss: 1.4688	 Generator Loss: 0.5680
Epoch: 1/10	Step: 2300	 Discriminator Loss: 1.3178	 Generator Loss: 0.7057
Epoch: 1/10	Step: 2320	 Discriminator Loss: 1.4047	 Generator Loss: 0.8789
Epoch: 1/10	Step: 2340	 Discriminator Loss: 1.3541	 Generator Loss: 0.7319
Epoch: 1/10	Step: 2360	 Discriminator Loss: 1.6755	 Generator Loss: 0.3575
Epoch: 1/10	Step: 2380	 Discriminator Loss: 1.6449	 Generator Loss: 0.3674
Epoch: 1/10	Step: 2400	 Discriminator Loss: 1.4169	 Generator Loss: 0.5970
Epoch: 1/10	Step: 2420	 Discriminator Loss: 1.6598	 Generator Loss: 0.4040
Epoch: 1/10	Step: 2440	 Discriminator Loss: 1.3952	 Generator Loss: 0.8401
Epoch: 1/10	Step: 2460	 Discriminator Loss: 1.4055	 Generator Loss: 0.7451
Epoch: 1/10	Step: 2480	 Discriminator Loss: 1.5615	 Generator Loss: 0.4131
Epoch: 1/10	Step: 2500	 Discriminator Loss: 1.5747	 Generator Loss: 0.4376

Epoch: 1/10	Step: 2520	 Discriminator Loss: 1.3336	 Generator Loss: 0.7020
Epoch: 1/10	Step: 2540	 Discriminator Loss: 2.2237	 Generator Loss: 1.6102
Epoch: 1/10	Step: 2560	 Discriminator Loss: 1.3870	 Generator Loss: 0.9877
Epoch: 1/10	Step: 2580	 Discriminator Loss: 1.3994	 Generator Loss: 1.0112
Epoch: 1/10	Step: 2600	 Discriminator Loss: 1.4256	 Generator Loss: 0.5909
Epoch: 1/10	Step: 2620	 Discriminator Loss: 1.4364	 Generator Loss: 0.6123
Epoch: 1/10	Step: 2640	 Discriminator Loss: 1.4646	 Generator Loss: 0.5271
Epoch: 1/10	Step: 2660	 Discriminator Loss: 1.4444	 Generator Loss: 0.9302
Epoch: 1/10	Step: 2680	 Discriminator Loss: 1.4172	 Generator Loss: 0.6145
Epoch: 1/10	Step: 2700	 Discriminator Loss: 1.3450	 Generator Loss: 0.6983
Epoch: 1/10	Step: 2720	 Discriminator Loss: 1.4725	 Generator Loss: 0.6149
Epoch: 1/10	Step: 2740	 Discriminator Loss: 1.2856	 Generator Loss: 0.6873
Epoch: 1/10	Step: 2760	 Discriminator Loss: 1.5201	 Generator Loss: 0.6281
Epoch: 1/10	Step: 2780	 Discriminator Loss: 1.4376	 Generator Loss: 0.6910
Epoch: 1/10	Step: 2800	 Discriminator Loss: 1.3778	 Generator Loss: 0.6576
Epoch: 1/10	Step: 2820	 Discriminator Loss: 1.4010	 Generator Loss: 0.6316
Epoch: 1/10	Step: 2840	 Discriminator Loss: 1.3690	 Generator Loss: 0.7091
Epoch: 1/10	Step: 2860	 Discriminator Loss: 1.3268	 Generator Loss: 0.8127
Epoch: 1/10	Step: 2880	 Discriminator Loss: 1.3693	 Generator Loss: 0.7476
Epoch: 1/10	Step: 2900	 Discriminator Loss: 1.4340	 Generator Loss: 0.7379
Epoch: 1/10	Step: 2920	 Discriminator Loss: 1.5574	 Generator Loss: 0.9873
Epoch: 1/10	Step: 2940	 Discriminator Loss: 1.3693	 Generator Loss: 0.8330
Epoch: 1/10	Step: 2960	 Discriminator Loss: 1.3244	 Generator Loss: 0.6714
Epoch: 1/10	Step: 2980	 Discriminator Loss: 1.5234	 Generator Loss: 0.5075
Epoch: 1/10	Step: 3000	 Discriminator Loss: 1.3136	 Generator Loss: 0.7032

Epoch: 1/10	Step: 3020	 Discriminator Loss: 1.5345	 Generator Loss: 0.4317
Epoch: 1/10	Step: 3040	 Discriminator Loss: 1.4819	 Generator Loss: 0.5383
Epoch: 1/10	Step: 3060	 Discriminator Loss: 1.4402	 Generator Loss: 0.7594
Epoch: 1/10	Step: 3080	 Discriminator Loss: 1.2434	 Generator Loss: 0.9213
Epoch: 1/10	Step: 3100	 Discriminator Loss: 1.4162	 Generator Loss: 0.6085
Epoch: 1/10	Step: 3120	 Discriminator Loss: 1.3164	 Generator Loss: 0.7414
Epoch: 1/10	Step: 3140	 Discriminator Loss: 1.2767	 Generator Loss: 0.8518
Epoch: 1/10	Step: 3160	 Discriminator Loss: 1.4281	 Generator Loss: 0.5888
Epoch: 2/10	Step: 3180	 Discriminator Loss: 1.3996	 Generator Loss: 0.8167
Epoch: 2/10	Step: 3200	 Discriminator Loss: 1.6501	 Generator Loss: 0.3782
Epoch: 2/10	Step: 3220	 Discriminator Loss: 1.4537	 Generator Loss: 0.9054
Epoch: 2/10	Step: 3240	 Discriminator Loss: 1.3059	 Generator Loss: 0.7350
Epoch: 2/10	Step: 3260	 Discriminator Loss: 1.3983	 Generator Loss: 1.1328
Epoch: 2/10	Step: 3280	 Discriminator Loss: 1.4680	 Generator Loss: 0.5132
Epoch: 2/10	Step: 3300	 Discriminator Loss: 1.4578	 Generator Loss: 0.5983
Epoch: 2/10	Step: 3320	 Discriminator Loss: 1.4482	 Generator Loss: 0.5456
Epoch: 2/10	Step: 3340	 Discriminator Loss: 1.3315	 Generator Loss: 0.8530
Epoch: 2/10	Step: 3360	 Discriminator Loss: 1.4852	 Generator Loss: 0.8314
Epoch: 2/10	Step: 3380	 Discriminator Loss: 1.4468	 Generator Loss: 0.5785
Epoch: 2/10	Step: 3400	 Discriminator Loss: 1.3755	 Generator Loss: 0.6240
Epoch: 2/10	Step: 3420	 Discriminator Loss: 1.2481	 Generator Loss: 0.9900
Epoch: 2/10	Step: 3440	 Discriminator Loss: 1.3889	 Generator Loss: 0.7670
Epoch: 2/10	Step: 3460	 Discriminator Loss: 1.5931	 Generator Loss: 0.4605
Epoch: 2/10	Step: 3480	 Discriminator Loss: 1.2523	 Generator Loss: 0.8909
Epoch: 2/10	Step: 3500	 Discriminator Loss: 1.3513	 Generator Loss: 0.6372

Epoch: 2/10	Step: 3520	 Discriminator Loss: 1.5059	 Generator Loss: 0.4871
Epoch: 2/10	Step: 3540	 Discriminator Loss: 1.4803	 Generator Loss: 0.5145
Epoch: 2/10	Step: 3560	 Discriminator Loss: 1.4063	 Generator Loss: 0.5568
Epoch: 2/10	Step: 3580	 Discriminator Loss: 1.4174	 Generator Loss: 0.5364
Epoch: 2/10	Step: 3600	 Discriminator Loss: 1.2902	 Generator Loss: 1.0374
Epoch: 2/10	Step: 3620	 Discriminator Loss: 1.4608	 Generator Loss: 0.9718
Epoch: 2/10	Step: 3640	 Discriminator Loss: 1.3675	 Generator Loss: 0.6967
Epoch: 2/10	Step: 3660	 Discriminator Loss: 1.4914	 Generator Loss: 0.5007
Epoch: 2/10	Step: 3680	 Discriminator Loss: 1.3674	 Generator Loss: 0.7496
Epoch: 2/10	Step: 3700	 Discriminator Loss: 1.5170	 Generator Loss: 0.5386
Epoch: 2/10	Step: 3720	 Discriminator Loss: 1.3668	 Generator Loss: 0.6969
Epoch: 2/10	Step: 3740	 Discriminator Loss: 1.3603	 Generator Loss: 0.7651
Epoch: 2/10	Step: 3760	 Discriminator Loss: 1.3547	 Generator Loss: 0.7408
Epoch: 2/10	Step: 3780	 Discriminator Loss: 1.5059	 Generator Loss: 0.4867
Epoch: 2/10	Step: 3800	 Discriminator Loss: 1.4624	 Generator Loss: 0.7626
Epoch: 2/10	Step: 3820	 Discriminator Loss: 1.4025	 Generator Loss: 0.6848
Epoch: 2/10	Step: 3840	 Discriminator Loss: 1.3154	 Generator Loss: 0.7576
Epoch: 2/10	Step: 3860	 Discriminator Loss: 1.3913	 Generator Loss: 0.7397
Epoch: 2/10	Step: 3880	 Discriminator Loss: 1.2762	 Generator Loss: 0.6971
Epoch: 2/10	Step: 3900	 Discriminator Loss: 1.3764	 Generator Loss: 0.7020
Epoch: 2/10	Step: 3920	 Discriminator Loss: 1.5473	 Generator Loss: 0.4367
Epoch: 2/10	Step: 3940	 Discriminator Loss: 1.4589	 Generator Loss: 0.5406
Epoch: 2/10	Step: 3960	 Discriminator Loss: 1.4011	 Generator Loss: 0.6639
Epoch: 2/10	Step: 3980	 Discriminator Loss: 1.5158	 Generator Loss: 0.5561
Epoch: 2/10	Step: 4000	 Discriminator Loss: 1.4092	 Generator Loss: 0.8370

Epoch: 2/10	Step: 4020	 Discriminator Loss: 1.4155	 Generator Loss: 0.6200
Epoch: 2/10	Step: 4040	 Discriminator Loss: 1.3794	 Generator Loss: 0.7964
Epoch: 2/10	Step: 4060	 Discriminator Loss: 1.4794	 Generator Loss: 0.7077
Epoch: 2/10	Step: 4080	 Discriminator Loss: 1.6325	 Generator Loss: 0.5045
Epoch: 2/10	Step: 4100	 Discriminator Loss: 1.4587	 Generator Loss: 0.4881
Epoch: 2/10	Step: 4120	 Discriminator Loss: 1.4389	 Generator Loss: 0.5515
Epoch: 2/10	Step: 4140	 Discriminator Loss: 1.3668	 Generator Loss: 0.6035
Epoch: 2/10	Step: 4160	 Discriminator Loss: 1.4059	 Generator Loss: 0.7834
Epoch: 2/10	Step: 4180	 Discriminator Loss: 1.5420	 Generator Loss: 0.4627
Epoch: 2/10	Step: 4200	 Discriminator Loss: 1.4423	 Generator Loss: 0.7255
Epoch: 2/10	Step: 4220	 Discriminator Loss: 1.4125	 Generator Loss: 0.7681
Epoch: 2/10	Step: 4240	 Discriminator Loss: 1.4406	 Generator Loss: 0.8161
Epoch: 2/10	Step: 4260	 Discriminator Loss: 1.4504	 Generator Loss: 0.7905
Epoch: 2/10	Step: 4280	 Discriminator Loss: 1.3602	 Generator Loss: 0.9260
Epoch: 2/10	Step: 4300	 Discriminator Loss: 1.5088	 Generator Loss: 0.5258
Epoch: 2/10	Step: 4320	 Discriminator Loss: 1.4703	 Generator Loss: 0.8564
Epoch: 2/10	Step: 4340	 Discriminator Loss: 1.5148	 Generator Loss: 0.4887
Epoch: 2/10	Step: 4360	 Discriminator Loss: 1.4858	 Generator Loss: 0.9431
Epoch: 2/10	Step: 4380	 Discriminator Loss: 1.4365	 Generator Loss: 0.6420
Epoch: 2/10	Step: 4400	 Discriminator Loss: 1.3755	 Generator Loss: 0.7190
Epoch: 2/10	Step: 4420	 Discriminator Loss: 1.4673	 Generator Loss: 0.6363
Epoch: 2/10	Step: 4440	 Discriminator Loss: 1.6052	 Generator Loss: 0.9288
Epoch: 2/10	Step: 4460	 Discriminator Loss: 1.4345	 Generator Loss: 0.6003
Epoch: 2/10	Step: 4480	 Discriminator Loss: 1.3843	 Generator Loss: 0.7293
Epoch: 2/10	Step: 4500	 Discriminator Loss: 1.3884	 Generator Loss: 0.6229

Epoch: 2/10	Step: 4520	 Discriminator Loss: 1.4263	 Generator Loss: 0.9199
Epoch: 2/10	Step: 4540	 Discriminator Loss: 1.4430	 Generator Loss: 0.9744
Epoch: 2/10	Step: 4560	 Discriminator Loss: 1.3608	 Generator Loss: 0.6312
Epoch: 2/10	Step: 4580	 Discriminator Loss: 1.3022	 Generator Loss: 0.9226
Epoch: 2/10	Step: 4600	 Discriminator Loss: 1.3655	 Generator Loss: 0.6699
Epoch: 2/10	Step: 4620	 Discriminator Loss: 1.4473	 Generator Loss: 0.5293
Epoch: 2/10	Step: 4640	 Discriminator Loss: 1.5900	 Generator Loss: 1.0348
Epoch: 2/10	Step: 4660	 Discriminator Loss: 1.1634	 Generator Loss: 1.1207
Epoch: 2/10	Step: 4680	 Discriminator Loss: 1.2902	 Generator Loss: 0.8838
Epoch: 2/10	Step: 4700	 Discriminator Loss: 1.4648	 Generator Loss: 0.7741
Epoch: 2/10	Step: 4720	 Discriminator Loss: 1.4008	 Generator Loss: 0.6272
Epoch: 2/10	Step: 4740	 Discriminator Loss: 1.3659	 Generator Loss: 0.6909
Epoch: 2/10	Step: 4760	 Discriminator Loss: 1.4338	 Generator Loss: 0.6306
Epoch: 2/10	Step: 4780	 Discriminator Loss: 1.6644	 Generator Loss: 0.8167
Epoch: 2/10	Step: 4800	 Discriminator Loss: 1.4471	 Generator Loss: 0.6700
Epoch: 2/10	Step: 4820	 Discriminator Loss: 1.4398	 Generator Loss: 0.6982
Epoch: 2/10	Step: 4840	 Discriminator Loss: 1.3833	 Generator Loss: 0.6688
Epoch: 2/10	Step: 4860	 Discriminator Loss: 1.4264	 Generator Loss: 0.6106
Epoch: 2/10	Step: 4880	 Discriminator Loss: 1.4271	 Generator Loss: 0.6758
Epoch: 2/10	Step: 4900	 Discriminator Loss: 1.3084	 Generator Loss: 0.8290
Epoch: 2/10	Step: 4920	 Discriminator Loss: 1.4665	 Generator Loss: 0.6980
Epoch: 2/10	Step: 4940	 Discriminator Loss: 1.4754	 Generator Loss: 0.5317
Epoch: 2/10	Step: 4960	 Discriminator Loss: 1.3773	 Generator Loss: 0.7428
Epoch: 2/10	Step: 4980	 Discriminator Loss: 1.3985	 Generator Loss: 0.7076
Epoch: 2/10	Step: 5000	 Discriminator Loss: 1.4832	 Generator Loss: 0.6758

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.